Voice signal collecting method and mobile terminal

Abstract

The invention provides a voice signal collecting method and a mobile terminal. The voice signal collecting method includes the following steps: collecting voice signals of a user, maintaining threshold value information corresponding to the user according to the voice signals, collecting voice input signals input by the user after a voice recognition mode is adopted, and judging a collecting starting point and a collecting stopping point in the voice input signals according to the threshold value information to recognize the voice input signals between the collecting starting point and the collecting stopping point. According to the embodiment, the corresponding threshold value information is maintained through the voice signals of the user, and therefore the threshold value information changes along changes of factors such as the user and the environment; the collecting starting point and the collecting stopping point in the determined voice input signals are judged to be accurate according to the threshold value information, effective voice data can be accurately divided, the recognition result accurate rate can be improved, the network transmission data volume can be reduced, and the user experience can be promoted.

Description

technical field [0001] The invention relates to the technical field of information processing, in particular to a voice signal collection method and a mobile terminal. Background technique [0002] In the speech recognition process, a client such as a mobile terminal needs to upload the collected speech data to the server. Since the amount of transmitted speech data is large, it is only necessary to selectively upload the speech data spoken by the user to be valid speech data. At present, the starting point and the ending point of the user's speech can be determined by setting a threshold, and judging the size of the speech energy and the threshold to determine the speech data of the user's speech. For example, if the speech energy is less than the threshold and the speech energy is greater than the threshold, it can be determined to start speaking. , the speech energy is greater than the threshold to the speech energy is less than the threshold and the state remains unchang...

AI Technical Summary

Problems solved by technology

[0003] The current problem is that the set threshold is not universal and is easily affected by various factors. It cannot accurately divide valid voice data, and the recognition results are not accurate, resulting in a large network transmission volume and poor user experience. For example, it is affected by user groups. Some people speak loudly, some people speak softly, and they are subject to the rigidity of the client system. Different hardware leads to different recording volumes. If the threshold is set too small, the background sound will be regarded as valid data, which will affect the recognition results. If the threshold is defined too Large, inaccurate monitoring of the time point of speaking, etc.

Images